[OE-core] [PATCH] xz: update to 5.1.2alpha

Koen Kooi koen at dominion.thruhere.net
Thu Jul 12 14:38:56 UTC 2012


Op 12 jul. 2012, om 15:42 heeft Andrei Gherzan het volgende geschreven:

> On Thu, Jul 12, 2012 at 3:56 PM, Koen Kooi <koen at dominion.thruhere.net> wrote:
> 
> Op 12 jul. 2012, om 14:26 heeft Andrei Gherzan het volgende geschreven:
> 
> > On Wed, Jul 11, 2012 at 11:53 PM, Saul Wold <sgw at linux.intel.com> wrote:
> > On 07/11/2012 07:10 AM, Koen Kooi wrote:
> > This contains bugfixes from the 5.0.3 and 5.0.4 releases
> >
> > Signed-off-by: Koen Kooi <koen at dominion.thruhere.net>
> > ---
> >   meta/recipes-extended/xz/{xz_5.1.1alpha.bb => xz_5.1.2alpha.bb} |    7 ++-----
> >   1 file changed, 2 insertions(+), 5 deletions(-)
> >   rename meta/recipes-extended/xz/{xz_5.1.1alpha.bb => xz_5.1.2alpha.bb} (86%)
> >
> > diff --git a/meta/recipes-extended/xz/xz_5.1.1alpha.bb b/meta/recipes-extended/xz/xz_5.1.2alpha.bb
> > similarity index 86%
> > rename from meta/recipes-extended/xz/xz_5.1.1alpha.bb
> > rename to meta/recipes-extended/xz/xz_5.1.2alpha.bb
> > index 7fa7177..9d68720 100644
> > --- a/meta/recipes-extended/xz/xz_5.1.1alpha.bb
> > +++ b/meta/recipes-extended/xz/xz_5.1.2alpha.bb
> > @@ -10,11 +10,8 @@ LIC_FILES_CHKSUM = "file://COPYING;md5=c475b6c7dca236740ace4bba553e8e1c \
> >                       file://lib/getopt.c;endline=23;md5=2069b0ee710572c03bb3114e4532cd84 "
> >
> >   SRC_URI = "http://tukaani.org/xz/xz-${PV}.tar.gz"
> > -
> > -SRC_URI[md5sum] = "bb24436fa12780808e1c142980484104"
> > -SRC_URI[sha256sum] = "54e59a83690a4a0ec88a7d7c3bdef90c6b196c892a93463863c71c24fe87951a"
> > -
> > -PR = "r0"
> > +SRC_URI[md5sum] = "9bad1e249537ce69b206815cf28ca87b"
> > +SRC_URI[sha256sum] = "70e792d2a67cfbb8f2dffd0feab6ca6e5a4a618d65070fb44a367629d1ba94e5"
> >
> >   inherit autotools gettext
> >
> >
> >
> > How stable is this alpha release?  We don't generally take up re-releases, correct?
> >
> >
> > We already have an alpha release... so another alpha shouldn't be a problem, right? .1a it's a good deal as it implements multithreaded compression and decompression.
> 
> After Andrei's patch for image_types.bbclass goes in I have a follow up patch to enable xz threads (and disable the -e -9) to give a nice speedup for xz images. On my aging core2quad compressing a 4gb image goes from ~22 minutes to ~6.5 minutes with -T4. On my 4c/8t i7 I can get up to a 6x speedup with -T6 or -T8, hyperthreading does a better job than I expected.
> 
>  
> I can confirm this. But still, we can hold -e -9 and just add -T4. I think people have 700mb ram to spare while compressing. 

It's 6.5GB of ram when I compress a 4GB image with -e -9 -T6, the -e is really, really bad



More information about the Openembedded-core mailing list